An intended interest point detection method provided for detecting an intended POI of a person from a plurality of first images and at least one second image includes obtaining a plurality of gaze information of the person in the plurality of first images; obtaining a location information of at least one POI in the at least one second image; determining energy value by the plurality of gaze information and the at least one POI; and inferring the intended POI by minimizing the energy value. The energy value is decreased when the gaze point location is correlated to a POI, and/or when the intended POI is consistent in contiguous time frames.
Legal claims defining the scope of protection, as filed with the USPTO.
1. An intended interest point detection method for detecting an intended point of interest (POI) of a person from a plurality of first images and at least one second image, the intended POI detection method comprising: obtaining a plurality of gaze information of the person in the plurality of first images, wherein the plurality of first images are consecutive frames captured by a first camera; obtaining a location information of at least one POI in the at least one second image; building a plurality of first layer nodes, wherein each first layer node stores values for one of the plurality of gaze information in a frame; building a plurality of second layer nodes, in which at least one second layer node is corresponded to each of the first layer node in time, wherein each of the plurality of second layer nodes represents a POI in a frame, and a value of the second layer node is a probability value indicating whether the person is attended to the represented POI or not; determining a spatial relation value by the first layer node and its time-corresponded at least one second layer node, wherein the spatial relation value is minimized when the gaze point location represented by the first layer node is correlated to the location of the POIs represented by the second layer nodes; determining a temporal relation value by a plurality of second layer nodes in different time frames, wherein the temporal relation value is minimized if the values of the second layer nodes representing the POIs in the consecutive frames are the same; determining an energy value by the plurality of gaze information and the at least one POI, wherein the energy value is decreased when the gaze point location is correlated to a POI, and/or when the intended POI is consistent in contiguous time frames; and minimizing the energy value composed by spatial relation and/or temporal relation and inferring the intended POI from the node corresponding to the maximum probability value of the second layer nodes in an image at which the energy value is minimized.
2. The intended interest point detection method of claim 1 , wherein the number of consecutive frames is adapted to a moving speed of the person, a moving speed of the at least one POI, and/or the frame rate of the first camera.
3. The intended interest point detection method of claim 1 , wherein the step of inferring the intended POI by minimizing the energy value comprises: finding values of the plurality of second layer nodes at which the energy value is minimized.
4. The intended interest point detection method of claim 1 , wherein the number of the second layer nodes are dynamically changed based on the number of POIs obtained in the at least one second image.
5. The intended interest point detection method of claim 1 , wherein the plurality of gaze information is a gaze point, a gaze area, or a probability distribution of a gaze region.
6. The intended interest point detection method of claim 1 , wherein the spatial relation between the plurality of gaze information and the at least one POI comprises a likelihood of the person's gaze point to the location of the at least one POI.
7. The intended interest point detection method of claim 1 , wherein the intended POI is either one of the at least one POI or none of the at least one POI.
8. The intended interest point detection method of claim 6 , wherein the likelihood is calculated based on a Gaussian distribution function.
9. An intended interest point detection system for detecting an intended point of interest (POI) of a person from a plurality of first images and at least one second image, the intended interest point detection system comprising: a gaze processing unit configured to obtain a plurality of gaze information of the person in the plurality of first images, wherein the plurality of first images are consecutive frames captured by a first camera; a scene processing unit configured to obtain a location information of at least one POI in the at least one second image; and an inference processing unit coupled to the gaze processing unit and the scene processing unit, and configured to receive the plurality of gaze information and the location information of the at least one POI, the inference processing unit executing instructions to perform operations comprising: building a plurality of first layer nodes, in which each first layer node stores values for one of the plurality of gaze information in a frame; building a plurality of second layer nodes, in which at least one second layer node is corresponded to each of the first layer node in time, each of the plurality of second layer nodes represents a POI in a frame, and a value of the second layer node is a probability value indicating whether the person is attended to the represented POI or not; determining a spatial relation value by the first layer node and its time-corresponded at least one second layer node, wherein the spatial relation value is minimized when the gaze point location represented by the first layer node is correlated to the location of the POIs represented by the second layer nodes; determining a temporal relation value by a plurality of second layer nodes in different time frames, wherein the temporal relation value is minimized if the values of the second layer nodes representing the POIs in the consecutive frames are the same; determining an energy value by the plurality of gaze information and the at least one POI; and minimizing the energy value composed by spatial relation and/or temporal relation, and inferring the intended POI from the node corresponding to the maximum probability value of the second layer nodes in an image at which the energy value is minimized; wherein the energy value is decreased when the gaze point location is correlated to a POI, and/or when the intended POI is consistent in contiguous time frames.
10. The intended interest point detection system of claim 9 , wherein the number of consecutive frames is adapted to a moving speed of the person, a moving speed of the at least one POI, and/or the frame rate of the first camera.
11. The intended interest point detection system of claim 9 , wherein the inference processing unit finds values of the plurality of second layer nodes at which the energy value is minimized.
12. The intended interest point detection system of claim 9 , wherein the number of the second layer nodes are dynamically changed based on the number of POIs obtained in the at least one second image.
13. The intended interest point detection system of claim 9 , wherein the plurality of gaze information is a gaze point, a gaze area, or a probability distribution of a gaze region.
14. The intended interest point detection system of claim 9 , wherein the spatial relation between the plurality of gaze information and the at least one POI comprises a likelihood of the person's gaze point to the location of the at least one POI.
15. The intended interest point detection system of claim 9 , wherein the intended POI is either one of the at least one POI or none of the at least one POI.
16. An intended interest point detection method for detecting an intended point of interest (POI) of a person in captured images, the intended interest point detection method comprising: obtaining a plurality of gaze information of the person in the captured images; obtaining a location information of at least one POI in the captured images; building a plurality of first layer nodes, wherein each first layer node stores values for one of the plurality of gaze information in a frame; building a plurality of second layer nodes, in which at least one second layer node is corresponded to each of the first layer node in time, wherein each of the plurality of second layer nodes represents a POI in a frame, and a value of the second layer node is a probability value indicating whether the person is attended to the represented POI or not; determining a spatial relation value by the first layer node and its time-corresponded at least one second layer node, wherein the spatial relation value is minimized when the gaze point location represented by the first layer node is correlated to the location of the POIs represented by the second layer nodes; determining a temporal relation value by a plurality of second layer nodes in different time frames, wherein the temporal relation value is minimized if the values of the second layer nodes representing the POIs in the consecutive frames are the same; determining an energy value by the plurality of gaze information and the at least one POI; and minimizing the energy value composed by spatial relation and/or temporal relation and inferring the intended POI from the node corresponding to the maximum probability value of the second layer nodes in an image at which the energy value is minimized; wherein the energy value is decreased when the gaze point location is correlated to a POI, and/or when the intended POI is consistent in contiguous time frames.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
November 2, 2018
November 24, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.